def test_OutputShape_1d(self):
X = np.random.randn(100, 3)
Y = X @ np.array([1, 2, 3]) - 2
self.assertEqual(len(Y.shape), 1)
solver = BatchCholeskySolver().partial_fit(X[::2], Y[::2])
Yh = solver.predict(X[1::2])
self.assertEqual(len(Yh.shape), 1)
def test_OutputShape_1d():
X = np.random.randn(100, 3)
Y = X @ np.array([1, 2, 3]) - 2
assert len(Y.shape) == 1
solver = BatchCholeskySolver().partial_fit(X[::2], Y[::2])
Yh = solver.predict(X[1::2])
assert len(Yh.shape) == 1
def test_SingleStepSolution(self):
X = np.random.randn(100, 3)
Y = X @ np.array([1, 2, 3]) - 2
solver = BatchCholeskySolver().fit(X[::2], Y[::2])
Yh = solver.predict(X[1::2])
assert_allclose(Y[1::2], Yh, rtol=1e-3)
assert_allclose(solver.coef_, np.array([1, 2, 3]), rtol=1e-3)
assert_allclose(solver.intercept_, -2)
def test_SingleStepSolution():
X = np.random.randn(100, 3)
Y = X @ np.array([1, 2, 3]) - 2
solver = BatchCholeskySolver().fit(X[::2], Y[::2])
Yh = solver.predict(X[1::2])
assert Y[1::2] == approx(Yh)
assert solver.coef_ == approx(np.array([1, 2, 3]), rel=1e-3)
assert solver.intercept_ == approx(-2)
def test_OutputShape_2d():
X = np.random.randn(100, 3)
W = np.array([[1, 4], [2, 5], [3, 6]])
Y = X @ W - 2
assert len(Y.shape) == 2
solver = BatchCholeskySolver().partial_fit(X[::2], Y[::2])
Yh = solver.predict(X[1::2])
assert len(Yh.shape) == 2
def test_OutputShape_2d(self):
X = np.random.randn(100, 3)
W = np.array([[1, 4],
[2, 5],
[3, 6]])
Y = X @ W - 2
self.assertEqual(len(Y.shape), 2)
solver = BatchCholeskySolver().partial_fit(X[::2], Y[::2])
Yh = solver.predict(X[1::2])
self.assertEqual(len(Yh.shape), 2)
def test_PartialFit_SeveralParts():
X = np.random.randn(100, 3)
W = np.array([[1, 4], [2, 5], [3, 6]])
Y = X @ W - 2
solver = BatchCholeskySolver()
# give 1st and 2nd output column separately, multiply by 2
# to get same results as if both columns were present both times
solver.partial_fit(X[::2], 2 * Y[::2] @ np.array([[1, 0], [0, 0]]))
solver.partial_fit(X[::2], 2 * Y[::2] @ np.array([[0, 0], [0, 1]]))
Yh = solver.predict(X[1::2])
assert Y[1::2] == approx(Yh, rel=1e-3)
assert solver.coef_ == approx(W, rel=1e-3)
assert solver.intercept_ == approx(np.array([-2, -2]))
def test_CholeskySolverSklearn_IsScikitLearnEstimator(self):
solver = BatchCholeskySolver()
check_estimator(solver)